Water vapor tendency associated with MJO represented in global reanalysis products

The Japan Meteorological Agency (JMA) recently released a new global reanalysis product, JRA-55, which was produced with the use of a JMA operational forecast model and a 4D-VAR data assimilation system. As a part of the activity evaluating the quality of JRA-55, this study examines variability in precipitable water vapor (PWV) and its tendency associated with the Madden-Julian Oscillation (MJO) to compare JRA-55 with JRA-25 (the previous version of Japanese reanalysis) and ERA Interim. Since the tendency of the analyzed PWV (analyzed tendency) is decomposed into tendency produced by the forecast model (model tendency) and modification by the assimilation system (analysis increment), this study focuses on the degree to which the analyzed tendency associated with the MJO is produced by the model tendency. The period examined is 16 years of 1989-2004 when a number of satellite observations were assimilated into the reanalysis product and thus its quality in the tropics is expected to be enough high.

For all of the three reanalysis products, intraseasonal-scale PWV anomalies associated with the boreal-winter (November-April) MJO propagate eastward in the eastern hemisphere along with precipitation anomalies, exhibiting quite similar longitudinal and temporal pattern with the SSM/I product. The amplitude of the PWV anomaly of the reanalysis products is smaller than that of the SSM/I product, though. The analyzed tendency of the PWV anomaly from the reanalysis products also exhibits quite similar spatiotemporal pattern with the SSM/I product, including areas where the analyzed tendency is statistically significant.

In contrast to the analyzed tendency, spatiotemporal patterns of the model tendency of the PWV anomaly are different among the three reanalysis products, with JRA-55 exhibiting the model tendency most similar to the analyzed tendency. The model tendency of PWV is caused by vertically-integrated horizontal convergence of moisture, surface evaporation, and precipitation. It is interesting that spatiotemporal patterns of these three variables are more similar among the reanalysis products compared to the model tendency.

The model tendency of JRA-55 is positive (negative) to the east (west) of the positive PWV anomalies, indicating that the forecast model tends to move the anomaly eastward. On the other hand, the model tendency is negative also over the positive PWV anomalies, which is compensated by positive analysis increments that are statistically significant. This suggests that the forecast model tends to damp the MJO signal reproduced by JRA-55.

This study also examines boreal-summer (May-October) intraseasonal variability and obtains similar results reproduced by JRA-55, while the similarity between the analyzed and model tendency is lesser than the boreal-winter MJO.